najoomi asli amil baba kala jadu expert rawalpindi bangladesh uk usa
Gp experyment
1. Motivation Experimental design Results
An experiment on temptation, self-commitment
and attitude towards paternalism
Michaª Krawczyk1 and Šukasz Wo¹ny2
1University of Warsaw
2Warsaw School of Economics
2016
3. Motivation Experimental design Results
Why this study?
• casual observations and empirical evidence stress importance
of temptation and self-control in explaining individual behavior
(see Angeletos, Laibson, Repetto, Tobacman, Weinberg
(2001); Ameriks, Caplin, Leahy, Tyler (2007)).
• present in
• consumption choices
• personal nance
• company-client relations
• insurance and nancial markets . . .
• large literature on behavioral aspects of contracting in the
principal-agent setting, namely on trust, fairness, reciprocity
etc.
• importantly: self control problems arise even when no moral
hazard is present.
4. Motivation Experimental design Results
Theoretical perspectives
• Models of quasi-hyperbolic discounting (Strotz, Laibson):
predict preference reversals, demand for commitment (when
choosing the ex ante bad thing expected)
• Models of costly self-control (Gul and Pesendorfer): make
similar predictions, except that the agent wishes to avoid also
temptations she correctly anticipates to resist (because it takes
eort to do so once she faces them)
• Dicult but important to distinguish, from both theoretical
and policy perspectives.
5. Motivation Experimental design Results
Self-commitment and paternalism
• both seem to be related, because someone's choice is
restricted, with his own benet in mind
• but is demand for both driven by analogous mechanisms?
• if a tempting object decreases utility as proposed by GP,
having it removed by somebody else makes a decision maker
better o, so she should be grateful.
• coupled with some mild tendency to positive reciprocation
(which is commonly found in experiments) should result in
willingness to reward such restrictions, even at a (low)
monetary cost to self.
• by contrast for a hyperbolic dicounter, removing the option she
currently craves is frustrating, will probably trigger negative
reciprocity
• thus observing if commitment and support for paternalism are
correlated seems interesting
6. Motivation Experimental design Results
Related experiments
• Casari (2009) and others (see Bitterly, 2014 for a review):
about one-third are self control types.
• Kataria et al. (2014): reactions to restrictions imposed on
their menu of gambles. Outcomes matter (hindsight bias).
• Le Lec and Tarroux (2015): binary comparisons between
menus of websites for self and another. Preferred larger menus
for others, restricted for self.
• Uhl (2013) paid for showing up for a session early in the
morning. Proportions of subjects willing to impose a
commitment on self and others were similar but no correlation
• in a survey of Danish students, Pendersen et al. (2014) found
some link between self-control and support for strong (such as
choice restrictions) but not weak (such as nudges) forms of
paternalism.
8. Motivation Experimental design Results
Pre-test
Which one would you choose... (1=certainly turkey, 2=rather t.,
3=hard to say, 4=rather burger 5=certainly burger)
stats | to eat to eat often for your
now in the future kids
---------+-------------------------------------
mean | 2.896 1.955 1.77
median | 2 2 1
-----------------------------------------------
N=135
9. Motivation Experimental design Results
Pre-test
Which dish is more... (1=certainly turkey etc.)
stats | eco healthy tasty tempting attr. satiating
--------+------------------------------------------------
mean | 2.007 1.333 3.333 3.622 3.311 3.67
median | 2 1 3 4 4 4
---------------------------------------------------------
n=135
10. Motivation Experimental design Results
Main study: decisions tasks
We invite subjects for a free lunch with opportunity to earn some
money and observe 4 choices:
• what menus (must eat burger; must eat turkey; your choice:
burger or turkey) subject assign to others;
• how they reward each of these menu choices another
participant made for them;
• which of the two dishes they pick on the spot (if given the
choice);
• whether they want to pre-commit to a choice of dish for a
future session.
11. Motivation Experimental design Results
Predictions
• GP theory rationalizes preference for commitment with some
self control V (T) V (T ∪ B) V (B)
• Strotz's overwhelming temptation rationalizes
V (T) V (T ∪ B) = V (B) while
• Kreps rationalizes preference for exibility
V (T ∪ B) V (T) V (B)
12. Motivation Experimental design Results
Predictions
• standard model: no commitment, no rewards
• standard with reciprocity: no commitment, rewards:
choicepreferredunpreferred
• hyperbolic discounting, no reciprocity: commitment (mostly to
turkey), no rewards
• hyperbolic discounting with reciprocity: commitment (mostly
to turkey), rewards: choicepreferredunpreferred
• costly self-control, no reciprocity: commitment (mostly to
turkey), no rewards
• costly self-control with reciprocity: commitment (mostly to
turkey), rewards: preferred ≥ [turkey: ] choice unpreferred
13. Motivation Experimental design Results
Treatments
four orthogonal manipulations
• order of dishes
• order of choices
• numbers in the example
• commitment before or after consumption
The last manipulation allowed investigating if willingness to commit
depends on visceral factors such as being hungry vs satiated.
16. Motivation Experimental design Results
Reaction to paternalism. Self-commitment
Variable | Obs Mean Std. Dev. Min Max
----------------+-----------------------------------------
reward_burger | 199 2.06 1.541 0 5
reward_turkey | 199 2.095 1.542 0 5
reward_both | 199 2.68 1.541 0 5
bonus f. choice | 199 -.0125 1.725 -5 5
• own preference main predictor of single-item rewards. 7/199
reward the unpreferred single-item menu higher than the
preferred. Bonus for choice not signicantly dierent from 0
• 20.6% choose commitment. It is strongly linked to lower
bonus for choice (p = 0.0187). The eect survives also when
controlling for strength of preference. Self-commits give sig.
neg. bonus for choice. This is driven by those who chose
turkey, rather than burger for future consumption.
17. Motivation Experimental design Results
Predictions revisited
• standard model: no commitment, no rewards
• standard with reciprocity: no commitment, rewards:
choicepreferredunpreferred
• hyperbolic discounting, no reciprocity: commitment (mostly to
turkey), no rewards
• hyperbolic discounting with reciprocity: commitment (mostly
to turkey), rewards: choicepreferredunpreferred
• costly self-control, no reciprocity: commitment (mostly to
turkey), no rewards
• costly self-control with reciprocity: commitment (mostly to
turkey), rewards: preferred ≥ [turkey: ] choice unpreferred
18. Motivation Experimental design Results
Self-commitment and paternalistic choices
| gave a choice
commitment | no yes | Total
-----------+----------------------+----------
no | 44 114 | 158
yes | 21 20 | 41
-----------+----------------------+----------
Total | 65 134 | 199
Pearson chi2(1) = 8.0844 Pr = 0.004
Again, this is robust to including controls, also condence in own
preference.
19. Motivation Experimental design Results
Other results
• no impact of manipulation of order of dishes, tasks or examples
• no impact of hunger (pre vs post) on willingness to commit
• no gender eects
• no eect of school, major (econ vs. not-econ)
• non-voters and those who voted for the right-wing,
conservative party (Law and Justice, PiS) more likely to
reward paternalistic choices (lower bonus for choice), more
likely to self-commit. PiS voters also more likely to restrict
others' choice
20. Motivation Experimental design Results
Summary
• We run a novel lab experiment with naturalistic stimuli
• We verify that temptation is relevant
• We observe non-trivial fraction of subjects self-committing,
choosing paternalistically for others and rewarding others'
patternalism
• these three tendencies are intercorrelated; overall, some
support for models of costly self-control
• some support for the claim that this is associated with
conservative political views